The determination of gene expression levels in tissues is of great importance for accurately diagnosing human disease and is increasingly used to determine a patient's course of treatment. Pharmacogenomic methods can identify patients likely to respond to a particular drug and can lead the way to new therapeutic approaches.
For example, thymidylate synthase (TS) is an integral enzyme in DNA biosynthesis where it catalyzes the reductive methylation of deoxyuridine monophosphate (dUMP) to deoxythymidine monophosphate (dTMP) and provides the only route for de novo synthesis of pyrimidine nucleotides within the cell (Johnston et al., 1995). Thymidylate synthase is a target for chemotherapeutic drugs, most commonly the antifolate agent 5-fluorouracil (5-FU). As the most effective single agent for the treatment of colon, head and neck and breast cancers, the primary action of 5-FU is to inhibit TS activity, resulting in depletion of intracellular thymine levels and subsequently leading to cell death.
Considerable variation in TS expression has been reported among clinical tumor specimens from both primary tumors (Johnston et al., 1995; Lenz et al., 1995) and metastases (Farrugia et aL, 1997; Leichmann et aL, 1997). In colorectal cancer, for example, the ratio of TS expression in tumor tissue relative to normal gastrointestinal mucosal tissue has ranged from 2 to 10 (Ardalan and Zang, 1996).
Thymidylate synthase is also known to have clinical importance in the development of tumor resistance, as demonstrated by studies that have shown acute induction of TS protein and an increase in TS enzyme levels in neoplastic cells after exposure to 5-FU (Spears et al. 1982; Swain et al. 1989). The ability of a tumor to acutely overexpress TS in response to cytotoxic agents such as 5-FU may play a role in the development of fluorouracil resistance. Previous studies have shown that the levels of TS protein directly correlate with the effectiveness of 5-FU therapy, that there is a direct correlation between protein and RNA expression (Jackman et al, 1985) and that TS expression is a powerful prognostic marker in colorectal and breast cancer (Jackman et al., 1985; Horikoshi et al, 1992).
In advanced metastatic disease, both high TS mRNA, quantified by RT-PCR, and high TS protein expression, have been shown to predict a poor response to fluoropyrimidine-based therapy for colorectal (Johnston et aL, 1995, Farrugia et al., 1997, Leichman et al., 1997), gastric (Lenz et al, 1995, Alexander et al., 1995), and head and neck (Johnston et al., 1997) cancers. A considerable overlap between responders and non-responders was often present in the low TS category, but patients with TS levels above the median were predominantly non-responders. The predictive value of TS overexpression may be further enhanced if combined with other molecular characteristics such as levels of dihydropyrimidine dehydrogenase (DPD) and thymidine phosphorylase (TP) expression, replication error positive (RER+) status (Kitchens and Berger 1997), and p53 status (Lenz et al., 1997). Studies to date that have evaluated the expression of TS in human tumors suggest that the ability to predict response and outcome based upon TS expression in human tumors may provide the opportunity in the future to select patients most likely to benefit from TS-directed therapy.
Until now, quantitative tissue gene expression studies including those of TS expression have been limited to reverse transcriptase polymerase chain reaction (RT-PCR) amplification of RNA from frozen tissue. However, most pathological samples are not prepared as frozen tissues, but are routinely formalin-fixed and paraffin-embedded (FFPE) to allow for histological analysis and for archival storage. Gene expression levels can be monitored semi-quantitatively and indirectly in such fixed and embedded samples by using immunohistochemical staining to monitor protein expression levels. Because paraffin-embedded samples are widely available, rapid and reliable methods are needed for the isolation of nucleic acids, particularly RNA, from such samples.
A number of techniques exist for the purification of RNA from biological samples, but none are reliable for isolation of RNA from FFPE samples. For example, Chomczynski (U.S. Pat. No. 5,346,994) describes a method for purifying RNA from tissues based on a liquid phase separation using phenol and guanidine isothiocyanate. A biological sample is homogenized in an aqueous solution of phenol and guanidine isothiocyanate and the homogenate thereafter mixed with chloroform. Following centrifugation, the homogenate separates into an organic phase, an interphase and an aqueous phase. Proteins are sequestered in the organic phase, DNA in the interphase, and RNA in the aqueous phase. RNA can be precipitated from the aqueous phase. This method does not provide for the reliable isolation of RNA from formalin-fixed paraffin-embedded tissue samples.
Other known techniques for isolating RNA typically utilize either guanidine salts or phenol extraction, as described for example in Sambrook, J. et al., (1989) at pp. 7.3-7.24, and in Ausubel, F. M. et al., (1994) at pp. 4.0.3-4.4.7. However, none of the known methods provide reproducible quantitative results in the isolation of RNA from paraffin-embedded tissue samples.
Techniques for the isolation of RNA from paraffin-embedded tissues are particularly needed for the study of gene expression in tumor tissues. Expression levels of certain receptors or enzymes can indicate the likelihood of success of a particular treatment.
Truly quantitative TS gene expression studies have been limited to RT-PCR from frozen tissue, whereas semi-quantitative monitoring of TS protein expression in archival pathological material fixed onto glass slides has been available via immunohistochemical staining. Because of limitations in isolating RNA from archival pathological material, quantitative techniques for measuring gene expression levels from such samples were heretofore unavailable.